A New Face Recognition Scheme with Renewable Templates

Under the increasing demand for more reliable security systems, there is a renewed interest in human identification based on biometric. Since the biometric templates stored in database directly have introduced a number of security and privacy threats, the study to biometric template protection is necessary. This paper provides a new face recognition scheme with renewable templates. In this scheme, a real value feature vector is derived from a 2D face image first, and it is transformed to a renewable form mainly based on Gray encoding, error-correction encoding and Shamir´s secret sharing technique. The experimental results shows that the protection scheme dose not degrade the performance of the original face recognition system. Comparing to the traditional face recognition scheme, the scheme provide higher security.